src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwLJSh_GeomP1P1_c.c

   1 /*
   2  * Note: this file was generated by the Gromacs c kernel generator.
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  * Copyright (c) 2001-2012, The GROMACS Development Team
   9  *
  10  * Gromacs is a library for molecular simulation and trajectory analysis,
  11  * written by Erik Lindahl, David van der Spoel, Berk Hess, and others - for
  12  * a full list of developers and information, check out http://www.gromacs.org
  13  *
  14  * This program is free software; you can redistribute it and/or modify it under
  15  * the terms of the GNU Lesser General Public License as published by the Free
  16  * Software Foundation; either version 2 of the License, or (at your option) any
  17  * later version.
  18  *
  19  * To help fund GROMACS development, we humbly ask that you cite
  20  * the papers people have written on it - you can find them on the website.
  21  */
  22 #ifdef HAVE_CONFIG_H
  23 #include <config.h>
  24 #endif
  25
  26 #include <math.h>
  27
  28 #include "../nb_kernel.h"
  29 #include "types/simple.h"
  30 #include "vec.h"
  31 #include "nrnb.h"
  32
  33 /*
  34  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJSh_GeomP1P1_VF_c
  35  * Electrostatics interaction: Ewald
  36  * VdW interaction:            LennardJones
  37  * Geometry:                   Particle-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecEwSh_VdwLJSh_GeomP1P1_VF_c
  42                     (t_nblist * gmx_restrict                nlist,
  43                      rvec * gmx_restrict                    xx,
  44                      rvec * gmx_restrict                    ff,
  45                      t_forcerec * gmx_restrict              fr,
  46                      t_mdatoms * gmx_restrict               mdatoms,
  47                      nb_kernel_data_t * gmx_restrict        kernel_data,
  48                      t_nrnb * gmx_restrict                  nrnb)
  49 {
  50     int              i_shift_offset,i_coord_offset,j_coord_offset;
  51     int              j_index_start,j_index_end;
  52     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  53     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  54     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  55     real             *shiftvec,*fshift,*x,*f;
  56     int              vdwioffset0;
  57     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  58     int              vdwjidx0;
  59     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  60     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  61     real             velec,felec,velecsum,facel,crf,krf,krf2;
  62     real             *charge;
  63     int              nvdwtype;
  64     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  65     int              *vdwtype;
  66     real             *vdwparam;
  67     int              ewitab;
  68     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  69     real             *ewtab;
  70
  71     x                = xx[0];
  72     f                = ff[0];
  73
  74     nri              = nlist->nri;
  75     iinr             = nlist->iinr;
  76     jindex           = nlist->jindex;
  77     jjnr             = nlist->jjnr;
  78     shiftidx         = nlist->shift;
  79     gid              = nlist->gid;
  80     shiftvec         = fr->shift_vec[0];
  81     fshift           = fr->fshift[0];
  82     facel            = fr->epsfac;
  83     charge           = mdatoms->chargeA;
  84     nvdwtype         = fr->ntype;
  85     vdwparam         = fr->nbfp;
  86     vdwtype          = mdatoms->typeA;
  87
  88     sh_ewald         = fr->ic->sh_ewald;
  89     ewtab            = fr->ic->tabq_coul_FDV0;
  90     ewtabscale       = fr->ic->tabq_scale;
  91     ewtabhalfspace   = 0.5/ewtabscale;
  92
  93     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
  94     rcutoff          = fr->rcoulomb;
  95     rcutoff2         = rcutoff*rcutoff;
  96
  97     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
  98     rvdw             = fr->rvdw;
  99
 100     outeriter        = 0;
 101     inneriter        = 0;
 102
 103     /* Start outer loop over neighborlists */
 104     for(iidx=0; iidx<nri; iidx++)
 105     {
 106         /* Load shift vector for this list */
 107         i_shift_offset   = DIM*shiftidx[iidx];
 108         shX              = shiftvec[i_shift_offset+XX];
 109         shY              = shiftvec[i_shift_offset+YY];
 110         shZ              = shiftvec[i_shift_offset+ZZ];
 111
 112         /* Load limits for loop over neighbors */
 113         j_index_start    = jindex[iidx];
 114         j_index_end      = jindex[iidx+1];
 115
 116         /* Get outer coordinate index */
 117         inr              = iinr[iidx];
 118         i_coord_offset   = DIM*inr;
 119
 120         /* Load i particle coords and add shift vector */
 121         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 122         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 123         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 124
 125         fix0             = 0.0;
 126         fiy0             = 0.0;
 127         fiz0             = 0.0;
 128
 129         /* Load parameters for i particles */
 130         iq0              = facel*charge[inr+0];
 131         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 132
 133         /* Reset potential sums */
 134         velecsum         = 0.0;
 135         vvdwsum          = 0.0;
 136
 137         /* Start inner kernel loop */
 138         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 139         {
 140             /* Get j neighbor index, and coordinate index */
 141             jnr              = jjnr[jidx];
 142             j_coord_offset   = DIM*jnr;
 143
 144             /* load j atom coordinates */
 145             jx0              = x[j_coord_offset+DIM*0+XX];
 146             jy0              = x[j_coord_offset+DIM*0+YY];
 147             jz0              = x[j_coord_offset+DIM*0+ZZ];
 148
 149             /* Calculate displacement vector */
 150             dx00             = ix0 - jx0;
 151             dy00             = iy0 - jy0;
 152             dz00             = iz0 - jz0;
 153
 154             /* Calculate squared distance and things based on it */
 155             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 156
 157             rinv00           = gmx_invsqrt(rsq00);
 158
 159             rinvsq00         = rinv00*rinv00;
 160
 161             /* Load parameters for j particles */
 162             jq0              = charge[jnr+0];
 163             vdwjidx0         = 2*vdwtype[jnr+0];
 164
 165             /**************************
 166              * CALCULATE INTERACTIONS *
 167              **************************/
 168
 169             if (rsq00<rcutoff2)
 170             {
 171
 172             r00              = rsq00*rinv00;
 173
 174             qq00             = iq0*jq0;
 175             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 176             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 177
 178             /* EWALD ELECTROSTATICS */
 179
 180             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 181             ewrt             = r00*ewtabscale;
 182             ewitab           = ewrt;
 183             eweps            = ewrt-ewitab;
 184             ewitab           = 4*ewitab;
 185             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 186             velec            = qq00*((rinv00-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 187             felec            = qq00*rinv00*(rinvsq00-felec);
 188
 189             /* LENNARD-JONES DISPERSION/REPULSION */
 190
 191             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 192             vvdw6            = c6_00*rinvsix;
 193             vvdw12           = c12_00*rinvsix*rinvsix;
 194             vvdw             = (vvdw12 - c12_00*sh_vdw_invrcut6*sh_vdw_invrcut6)*(1.0/12.0) - (vvdw6 - c6_00*sh_vdw_invrcut6)*(1.0/6.0);
 195             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 196
 197             /* Update potential sums from outer loop */
 198             velecsum        += velec;
 199             vvdwsum         += vvdw;
 200
 201             fscal            = felec+fvdw;
 202
 203             /* Calculate temporary vectorial force */
 204             tx               = fscal*dx00;
 205             ty               = fscal*dy00;
 206             tz               = fscal*dz00;
 207
 208             /* Update vectorial force */
 209             fix0            += tx;
 210             fiy0            += ty;
 211             fiz0            += tz;
 212             f[j_coord_offset+DIM*0+XX] -= tx;
 213             f[j_coord_offset+DIM*0+YY] -= ty;
 214             f[j_coord_offset+DIM*0+ZZ] -= tz;
 215
 216             }
 217
 218             /* Inner loop uses 59 flops */
 219         }
 220         /* End of innermost loop */
 221
 222         tx = ty = tz = 0;
 223         f[i_coord_offset+DIM*0+XX] += fix0;
 224         f[i_coord_offset+DIM*0+YY] += fiy0;
 225         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 226         tx                         += fix0;
 227         ty                         += fiy0;
 228         tz                         += fiz0;
 229         fshift[i_shift_offset+XX]  += tx;
 230         fshift[i_shift_offset+YY]  += ty;
 231         fshift[i_shift_offset+ZZ]  += tz;
 232
 233         ggid                        = gid[iidx];
 234         /* Update potential energies */
 235         kernel_data->energygrp_elec[ggid] += velecsum;
 236         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 237
 238         /* Increment number of inner iterations */
 239         inneriter                  += j_index_end - j_index_start;
 240
 241         /* Outer loop uses 15 flops */
 242     }
 243
 244     /* Increment number of outer iterations */
 245     outeriter        += nri;
 246
 247     /* Update outer/inner flops */
 248
 249     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*15 + inneriter*59);
 250 }
 251 /*
 252  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJSh_GeomP1P1_F_c
 253  * Electrostatics interaction: Ewald
 254  * VdW interaction:            LennardJones
 255  * Geometry:                   Particle-Particle
 256  * Calculate force/pot:        Force
 257  */
 258 void
 259 nb_kernel_ElecEwSh_VdwLJSh_GeomP1P1_F_c
 260                     (t_nblist * gmx_restrict                nlist,
 261                      rvec * gmx_restrict                    xx,
 262                      rvec * gmx_restrict                    ff,
 263                      t_forcerec * gmx_restrict              fr,
 264                      t_mdatoms * gmx_restrict               mdatoms,
 265                      nb_kernel_data_t * gmx_restrict        kernel_data,
 266                      t_nrnb * gmx_restrict                  nrnb)
 267 {
 268     int              i_shift_offset,i_coord_offset,j_coord_offset;
 269     int              j_index_start,j_index_end;
 270     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 271     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 272     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 273     real             *shiftvec,*fshift,*x,*f;
 274     int              vdwioffset0;
 275     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 276     int              vdwjidx0;
 277     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 278     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 279     real             velec,felec,velecsum,facel,crf,krf,krf2;
 280     real             *charge;
 281     int              nvdwtype;
 282     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 283     int              *vdwtype;
 284     real             *vdwparam;
 285     int              ewitab;
 286     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 287     real             *ewtab;
 288
 289     x                = xx[0];
 290     f                = ff[0];
 291
 292     nri              = nlist->nri;
 293     iinr             = nlist->iinr;
 294     jindex           = nlist->jindex;
 295     jjnr             = nlist->jjnr;
 296     shiftidx         = nlist->shift;
 297     gid              = nlist->gid;
 298     shiftvec         = fr->shift_vec[0];
 299     fshift           = fr->fshift[0];
 300     facel            = fr->epsfac;
 301     charge           = mdatoms->chargeA;
 302     nvdwtype         = fr->ntype;
 303     vdwparam         = fr->nbfp;
 304     vdwtype          = mdatoms->typeA;
 305
 306     sh_ewald         = fr->ic->sh_ewald;
 307     ewtab            = fr->ic->tabq_coul_F;
 308     ewtabscale       = fr->ic->tabq_scale;
 309     ewtabhalfspace   = 0.5/ewtabscale;
 310
 311     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 312     rcutoff          = fr->rcoulomb;
 313     rcutoff2         = rcutoff*rcutoff;
 314
 315     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 316     rvdw             = fr->rvdw;
 317
 318     outeriter        = 0;
 319     inneriter        = 0;
 320
 321     /* Start outer loop over neighborlists */
 322     for(iidx=0; iidx<nri; iidx++)
 323     {
 324         /* Load shift vector for this list */
 325         i_shift_offset   = DIM*shiftidx[iidx];
 326         shX              = shiftvec[i_shift_offset+XX];
 327         shY              = shiftvec[i_shift_offset+YY];
 328         shZ              = shiftvec[i_shift_offset+ZZ];
 329
 330         /* Load limits for loop over neighbors */
 331         j_index_start    = jindex[iidx];
 332         j_index_end      = jindex[iidx+1];
 333
 334         /* Get outer coordinate index */
 335         inr              = iinr[iidx];
 336         i_coord_offset   = DIM*inr;
 337
 338         /* Load i particle coords and add shift vector */
 339         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 340         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 341         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 342
 343         fix0             = 0.0;
 344         fiy0             = 0.0;
 345         fiz0             = 0.0;
 346
 347         /* Load parameters for i particles */
 348         iq0              = facel*charge[inr+0];
 349         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 350
 351         /* Start inner kernel loop */
 352         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 353         {
 354             /* Get j neighbor index, and coordinate index */
 355             jnr              = jjnr[jidx];
 356             j_coord_offset   = DIM*jnr;
 357
 358             /* load j atom coordinates */
 359             jx0              = x[j_coord_offset+DIM*0+XX];
 360             jy0              = x[j_coord_offset+DIM*0+YY];
 361             jz0              = x[j_coord_offset+DIM*0+ZZ];
 362
 363             /* Calculate displacement vector */
 364             dx00             = ix0 - jx0;
 365             dy00             = iy0 - jy0;
 366             dz00             = iz0 - jz0;
 367
 368             /* Calculate squared distance and things based on it */
 369             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 370
 371             rinv00           = gmx_invsqrt(rsq00);
 372
 373             rinvsq00         = rinv00*rinv00;
 374
 375             /* Load parameters for j particles */
 376             jq0              = charge[jnr+0];
 377             vdwjidx0         = 2*vdwtype[jnr+0];
 378
 379             /**************************
 380              * CALCULATE INTERACTIONS *
 381              **************************/
 382
 383             if (rsq00<rcutoff2)
 384             {
 385
 386             r00              = rsq00*rinv00;
 387
 388             qq00             = iq0*jq0;
 389             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 390             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 391
 392             /* EWALD ELECTROSTATICS */
 393
 394             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 395             ewrt             = r00*ewtabscale;
 396             ewitab           = ewrt;
 397             eweps            = ewrt-ewitab;
 398             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 399             felec            = qq00*rinv00*(rinvsq00-felec);
 400
 401             /* LENNARD-JONES DISPERSION/REPULSION */
 402
 403             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 404             fvdw             = (c12_00*rinvsix-c6_00)*rinvsix*rinvsq00;
 405
 406             fscal            = felec+fvdw;
 407
 408             /* Calculate temporary vectorial force */
 409             tx               = fscal*dx00;
 410             ty               = fscal*dy00;
 411             tz               = fscal*dz00;
 412
 413             /* Update vectorial force */
 414             fix0            += tx;
 415             fiy0            += ty;
 416             fiz0            += tz;
 417             f[j_coord_offset+DIM*0+XX] -= tx;
 418             f[j_coord_offset+DIM*0+YY] -= ty;
 419             f[j_coord_offset+DIM*0+ZZ] -= tz;
 420
 421             }
 422
 423             /* Inner loop uses 41 flops */
 424         }
 425         /* End of innermost loop */
 426
 427         tx = ty = tz = 0;
 428         f[i_coord_offset+DIM*0+XX] += fix0;
 429         f[i_coord_offset+DIM*0+YY] += fiy0;
 430         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 431         tx                         += fix0;
 432         ty                         += fiy0;
 433         tz                         += fiz0;
 434         fshift[i_shift_offset+XX]  += tx;
 435         fshift[i_shift_offset+YY]  += ty;
 436         fshift[i_shift_offset+ZZ]  += tz;
 437
 438         /* Increment number of inner iterations */
 439         inneriter                  += j_index_end - j_index_start;
 440
 441         /* Outer loop uses 13 flops */
 442     }
 443
 444     /* Increment number of outer iterations */
 445     outeriter        += nri;
 446
 447     /* Update outer/inner flops */
 448
 449     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*13 + inneriter*41);
 450 }